Rock-Paper-Scissors is a classic random game, isn't it? No, most people play with a, more or less conscious, tactic.

Our game-engine uses a neural network to predict the move of the opponent. 
A game is played until any player has 10 wins. Statistics from previous games shows that a human player will win in 50 % of the first games. If the player continues for a second game the computer will win 73 % of these.

This game was made by Henrik Engström & Lars Niklasson University of Skövde, Sweden. 

More details on the system can be found in this paper presented at the International Conference on Application and Development of Computer Games in the 21st Century 2001.